基于LOD控制与内外存调度的大型三维点云数据绘制

Rendering of Huge Point-Sampled Geometry Based on LOD Control and Out-of-Core Techniques

通过结合基于视点的细节层次(level-of-detail,LOD)控制技术和内外存调度的数据控制策略,实现大型三维点云数据在一般配置PC机上的实时交互浏览.首先将输入点云分为大小相等的若干块,然后对每块数据分别建立误差控制下的多分辨率数据结构,并进行内外存分配.在交互绘制中,通过用户视点来确定当前的感兴趣区域,以控制模型表面的细节层次分布.该算法不但可以实现大型点云数据的实时交互绘制,而且可有效地提高一般点云数据绘制时的内存使用效率.

A new algorithm for interactive rendering of huge point-sampled geometry based on viewdependent level-of-detail （LOD） and out-of-core techniques is presented. At first, point clouds are partitioned into small blocks with the same size, and points in each block are reorganized into errorcontrolled LODs. After that, these LOD data are distributed into main and external memories for balancing system loads. In the interactive rendering, viewers can control the resolution distribution of points on the whole surfaces by defining current view parameters and using a paging algorithm. Experimental results show that our approach can not only realize real-time rendering for huge datasets, but also is able to improve the usage of main memory for general point clouds.